Overview

The CCHA power analysis evaluates the effect of reducing sampling effort for vegetation transects at all long-term monitoring sites in Tampa Bay. Currently, sampling is conducted every half meter and the power analysis evaluates the effect of reduced effort up to sampling every ten meters, or a 95% reduction from the current effort. Scenarios evaluating more than 50% (1 meter sampling) or 66% (1.5 meter sampling) reduction in effort are unrealistic, but still useful to understand the effects on quantitative measures of habitat. These measures include:

  1. Total species richness at a site
  2. Total species richness at a site by vegetation zone
  3. Species frequency occurrence at a site
  4. Vegetation zone distances by site
  5. Elevation range at which 95% of key species occur (e.g., mangroves)

The relative reduction in effort associated with each half-meter increase along a transect is shown below.

Sub-sampling scheme

All existing CCHA vegetation surveys were sub-sampled from the existing effort of sample plots every half meter. The plot below shows an example of the sub-sampling scheme, where effort was reduced in increments of a half meter, starting from the complete survey to an upper limit of sampling every ten meters. For simplicity, the plot shows sub-sampling up to every three meters for a hypothetical 30 meter transect (transect exceed 100 meters at all site).

The existing surveys were sub-sampled at the specified meter interval for every unique subset (or replicate) that was possible. For example, two unique replicates can be created with sub-sampling every one meter, three every 1.5 meters, etc.

The large red points show which of the existing survey points were sub-sampled for the specified sub-sample distance. For each site and sample year, the vegetation transect was sub-sampled following the scheme below (sub-sampling up to ten meters) and relevant habitat summaries were estimated. The top row represents the complete transect, or the “truth”.

Note: Although it appears that the entire original transect is sampled across all replicates, the results from each replicate are averaged as independent samples appropriate for the selected effort.

Summaries

Richness estimates

These plots show the effect of down-sampling on richness estimates at each site, grouped by year of sampling. The top plot shows points as the average estimate at each down-sampled survey for the sampling distance shown on the x-axis. The lines are polynomial smooths to show the trend. The size of each point is in proportion to the variance of the species richness estimate for each random sub-sample at the specified level of sampling. Facets are arranged based on the greatest reduction in species richness as a percentage of the total.

These plots are similar to the above, but richness is scaled as a percentage of the total at full sample effort. This allows for a comparison of a reduction in the estimate independent of the overall species richness. That is, the sensitivity of the richness estimate at a site may vary depending on overall species richness.

There is some evidence that a greater reduction in species richness is expected at sites with higher richness, although the model is insignificant.

Richness estimates by zone

These plots show richness estimates for different zones with a reduction of sample effort. The lines show the estimated reduction in the richness estimate for each site and the thick line is the average trend across all sites for each zone (points are not shown to reduce clutter). The panels are arranged based on the greatest percent reduction in richness from full to minimum effort across all sites in a zone.

This plot is the same as above, except richness estimates are scaled as a percentage of the total. This allows for a comparison of a reduction in the estimate independent of the overall species richness.

This plot combines results from the previous. The lines are each solid black line from the above plot as the average reduction in the richness estimates for all sites in each zone. The line thickness is in proportion to the average total species richness for all sites in a zone at full sampling effort.

There is strong evidence that a greater reduction in species richness is expected at zones with higher richness.

Frequency occurrence estimates

The plot below shows the estimated change in frequency occurrence with reduced sample effort at each site and sample year. Each line shows a different species (labels are not included to reduce plot clutter). The points are sized in proportion to the variance across the random subsamples at the specified level of effort.

Variance of frequency occurrence with each random subsample shows an increase with a reduction in sample effort. The increase in the variance of frequency occurrence depends on the actual frequency occurrence estimate with complete sample effort. The model below shows that species with higher frequency occurrence will have greater variation in the estimate with reductions in sample effort. The x-axis shows the initial frequency occurrence and the y-axis shows the final variance at 10 m sampling.

Zone counts and distances

Estimates of the counts and distances for each zone along the transects are shown below for reductions in sample effort. The first plot shows the number of unique zones identified with down-sampling. All zones are identified at each site for all levels of sample effort.

The estimated length of each zone at each site is shown below with reductions in sample effort. Each line represents a unique zone. Estimated zone lengths do not change with reduced effort. Points are sized by the variance of the estimates across the random subsamples at level of sample effort. Variance (or uncertainty) in the zone distances increases with reduced effort.

The variance (or uncertainy) of the distance estimates increases with reductions in effort, but the increase is not linear. An example for Big Bend - TECO is shown below. Although variance increase across the zones, the variance estimates fall to zero when the actual zone distance (facet label shown in meters) is divisible by the sample interval with no remainder. Points of maximum variance are between intervals with no remainder.

Detailed results for each site:

Conclusions

  • Richness estimates for a site are reduced with lower sample effort
  • The amount of reduction in the richness estimated was weakly associated with total species richness at a site
  • Richness estimates for a zone are reduced with lower sample effort
  • Richness estimates will be systematically lower with reduced effort for zones with overall higher richness. These include zones…
  • Frequency occurrence estimates for individual species will not be systematically different at lower sample effort, but the estimates are imprecise
  • Species with higher frequency occurrence will have less precise estimates with reduced effort compared to those with lower frequency occurrence
  • Zone identification at each site is not affected by sample effort, although distance estimates decrease slightly.
  • Precision of the distance estimates for each zone decreases with reductions in sample effort, but the decrease is not linear and is affected by the actual zone length and sample interval